SolrIndexTask::doReindex() - Code Metrics - Inspection of "Remove the Facet returned test for now, it seems u..." - Firesphere/silverstripe-solr-search - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — hans/cow-is-animal ( cb2f8e...10edfb )

by Simon

created 2019-12-03 05:05 UTC

SolrIndexTask::doReindex() A

↳ Parent: SolrIndexTask

Complexity

Conditions	5
Paths	6

Size

Total Lines	14
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	7
CRAP Score	5.2742

Importance

Changes	11
Bugs	2	Features	0

Metric	Value
cc	5
eloc	8
c	11
b	2
f	0
nc	6
nop	4
dl	0
loc	14
ccs	7
cts	9
cp	0.7778
crap	5.2742
rs	9.6111

<?php


namespace Firesphere\SolrSearch\Tasks;

use Exception;
use Firesphere\SolrSearch\Factories\DocumentFactory;
use Firesphere\SolrSearch\Helpers\SolrLogger;
use Firesphere\SolrSearch\Indexes\BaseIndex;
use Firesphere\SolrSearch\Services\SolrCoreService;
use Firesphere\SolrSearch\States\SiteState;
use Firesphere\SolrSearch\Traits\LoggerTrait;
use GuzzleHttp\Exception\GuzzleException;
use Psr\Log\LoggerInterface;
use ReflectionException;
use SilverStripe\Control\Controller;
use SilverStripe\Control\Director;
use SilverStripe\Control\HTTPRequest;
use SilverStripe\Core\Injector\Injector;
use SilverStripe\Dev\BuildTask;
use SilverStripe\ORM\ArrayList;
use SilverStripe\ORM\DataList;
use SilverStripe\ORM\DataObject;
use SilverStripe\ORM\DB;
use SilverStripe\ORM\ValidationException;
use SilverStripe\Versioned\Versioned;

/**
 * Class SolrIndexTask
 *
 * @description Index items to Solr through a tasks
 * @package Firesphere\SolrSearch\Tasks
 */
class SolrIndexTask extends BuildTask
{
    use LoggerTrait;
    /**
     * URLSegment of this task
     *
     * @var string
     */
    private static $segment = 'SolrIndexTask';
    /**
     * Store the current states for all instances of SiteState
     *
     * @var array
     */
    public $currentStates;
    /**
     * My name
     *
     * @var string
     */
    protected $title = 'Solr Index update';
    /**
     * What do I do?
     *
     * @var string
     */
    protected $description = 'Add or update documents to an existing Solr core.';
    /**
     * Debug mode enabled, default false
     *
     * @var bool
     */
    protected $debug = false;
    /**
     * Singleton of {@link SolrCoreService}
     *
     * @var SolrCoreService
     */
    protected $service;

    /**
     * @var int
     */
    protected $batchLength = 1;

    /**
     * SolrIndexTask constructor. Sets up the document factory
     *
     * @throws ReflectionException
     */
    public function __construct()
    {
        parent::__construct();
        // Only index live items.
        // The old FTS module also indexed Draft items. This is unnecessary
        Versioned::set_reading_mode(Versioned::DEFAULT_MODE);
        // If versioned is needed, a separate Versioned Search module is required
        $this->setService(Injector::inst()->get(SolrCoreService::class));
        $this->setLogger(Injector::inst()->get(LoggerInterface::class));
        $this->setDebug(Director::isDev() || Director::is_cli());
        $currentStates = SiteState::currentStates();
        SiteState::setDefaultStates($currentStates);
    }

    /**
     * Set the {@link SolrCoreService}
     *
     * @param SolrCoreService $service
     * @return SolrIndexTask
     */
    public function setService(SolrCoreService $service): SolrIndexTask
    {
        $this->service = $service;

        return $this;
    }

    /**
     * Set the debug mode
     *
     * @param bool $debug
     * @return SolrIndexTask
     */
    public function setDebug(bool $debug): SolrIndexTask
    {
        $this->debug = $debug;

        return $this;
    }

    /**
     * Implement this method in the task subclass to
     * execute via the TaskRunner
     *
     * @param HTTPRequest $request
     * @return int|bool
     * @throws Exception
     * @throws GuzzleException
     */
    public function run($request)
    {
        $start = time();
        $this->getLogger()->info(date('Y-m-d H:i:s'));
        list($vars, $group, $isGroup) = $this->taskSetup($request);
        $groups = 0;
        $indexes = $this->service->getValidIndexes($request->getVar('index'));

        foreach ($indexes as $indexName) {
            /** @var BaseIndex $index */
            $index = Injector::inst()->get($indexName, false);

            $indexClasses = $index->getClasses();
            $classes = $this->getClasses($vars, $indexClasses);
            if (!count($classes)) {
                continue;
            }

            $this->clearIndex($vars, $index);

            $groups = $this->indexClassForIndex($classes, $isGroup, $index, $group);
        }

        $this->getLogger()->info(date('Y-m-d H:i:s'));
        $this->getLogger()->info(sprintf('Time taken: %s minutes', (time() - $start) / 60));

        return $groups;
    }

    /**
     * Set up the requirements for this task
     *
     * @param HTTPRequest $request
     * @return array
     */
    protected function taskSetup($request): array
    {
        $vars = $request->getVars();
        $this->debug = $this->debug || isset($vars['debug']);
        $group = $vars['group'] ?? 0;
        $start = $vars['start'] ?? 0;
        $group = ($start > $group) ? $start : $group;
        $isGroup = isset($vars['group']);

        return [$vars, $group, $isGroup];
    }

    /**
     * get the classes to run for this task execution
     *
     * @param $vars
     * @param array $classes
     * @return bool|array
     */
    protected function getClasses($vars, array $classes): array
    {
        if (isset($vars['class'])) {
            return array_intersect($classes, [$vars['class']]);
        }

        return $classes;
    }

    /**
     * Clear the given index if a full re-index is needed
     *
     * @param $vars
     * @param BaseIndex $index
     * @throws Exception
     */
    public function clearIndex($vars, BaseIndex $index)
    {
        if (!empty($vars['clear'])) {
            $this->getLogger()->info(sprintf('Clearing index %s', $index->getIndexName()));
            $this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $index);
        }
    }

    /**
     * Index the classes for a specific index
     *
     * @param $classes
     * @param $isGroup
     * @param BaseIndex $index
     * @param $group
     * @return int
     * @throws Exception
     * @throws GuzzleException
     */
    protected function indexClassForIndex($classes, $isGroup, BaseIndex $index, $group): int
    {
        $groups = 0;
        foreach ($classes as $class) {
            $groups = $this->indexClass($isGroup, $class, $index, $group);
        }

        return $groups;
    }

    /**
     * Index a single class for a given index. {@link static::indexClassForIndex()}
     *
     * @param bool $isGroup
     * @param string $class
     * @param BaseIndex $index
     * @param int $group
     * @return int
     * @throws GuzzleException
     * @throws ValidationException
     */
    private function indexClass($isGroup, $class, BaseIndex $index, int $group): int
    {
        $this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName()));
        $this->batchLength = DocumentFactory::config()->get('batchLength');
        $totalGroups = (int)ceil($class::get()->count() / $this->batchLength);
        $cores = SolrCoreService::config()->get('cpucores') ?: 1;
        $groups = $isGroup ? ($group + $cores - 1) : $totalGroups;
        $this->getLogger()->info(sprintf('Total groups %s', $totalGroups));
        do { // Run from oldest to newest
            try {
                // The unittest param is from phpunit.xml.dist, meant to bypass the exit(0) call
                if (function_exists('pcntl_fork') &&
                    !Controller::curr()->getRequest()->getVar('unittest')
                ) {
                    $group = $this->spawnChildren($class, $index, $group, $cores, $groups);
                } else {
                    $this->doReindex($group, $class, $index);
                }
            } catch (Exception $error) {
                $this->logException($index->getIndexName(), $group, $error);
                $group++;
                continue;
            }
            $group++;
        } while ($group <= $groups);

        return $totalGroups;
    }

    /**
     * For each core, spawn a child process that will handle a separate group.
     * This speeds up indexing through CLI massively.
     *
     * @param string $class
     * @param BaseIndex $index
     * @param int $group
     * @param int $cores
     * @param int $groups
     * @return int
     * @throws Exception
     */
    private function spawnChildren($class, BaseIndex $index, int $group, int $cores, int $groups): int
    {
        $start = $group;
        $pids = [];
        // for each core, start a grouped indexing
        for ($i = 0; $i < $cores; $i++) {
            $start = $group + $i;
            if ($start < $groups) {
                $pid = pcntl_fork();
                // PID needs to be pushed before anything else, for some reason
                $pids[$i] = $pid;
                $config = DB::getConfig();
                DB::connect($config);
                if (!$pid) {
                    $this->doReindex($start, $class, $index, true);
                }
            }
        }
        // Wait for each child to finish
        foreach ($pids as $key => $pid) {
            pcntl_waitpid($pid, $status);
            if ($status === 0) {
                unset($pids[$key]);
            }
        }
        $commit = $index->getClient()->createUpdate();
        $commit->addCommit();

        $index->getClient()->update($commit);

        return $start;
    }

    /**
     * Reindex the given group, for each state
     *
     * @param int $group
     * @param string $class
     * @param BaseIndex $index
     * @param bool $pcntl
     * @throws Exception
     */
    private function doReindex($group, $class, BaseIndex $index, $pcntl = false)
    {
        foreach (SiteState::getStates() as $state) {
            if ($state !== 'default' && !empty($state)) {
                SiteState::withState($state);
            }
            $this->stateReindex($group, $class, $index);
        }

        SiteState::withState(SiteState::DEFAULT_STATE);
        $this->getLogger()->info(sprintf('Indexed group %s', $group));

        if ($pcntl) {
            exit(0);

        }
    }

    /**
     * Index a group of a class for a specific state and index
     *
     * @param $group
     * @param $class
     * @param BaseIndex $index
     * @throws Exception
     */
    private function stateReindex($group, $class, BaseIndex $index): void
    {
        // Generate filtered list of local records
        $baseClass = DataObject::getSchema()->baseDataClass($class);
        /** @var DataList|DataObject[] $items */
        $items = DataObject::get($baseClass)
            ->sort('ID ASC')
            ->limit($this->batchLength, ($group * $this->batchLength));
        if ($items->count()) {
            $this->updateIndex($index, $items);
        }
    }

    /**
     * Execute the update on the client
     *
     * @param BaseIndex $index
     * @param $items
     * @throws Exception
     */
    private function updateIndex(BaseIndex $index, $items): void
    {
        $client = $index->getClient();
        $update = $client->createUpdate();
        $this->service->setInDebugMode($this->debug);
        $this->service->updateIndex($index, $items, $update);
        $client->update($update);
    }

    /**
     * Log an exception if it happens. Most are catched, these logs are for the developers
     * to identify problems and fix them.
     *
     * @param string $index
     * @param int $group
     * @param Exception $exception
     * @throws GuzzleException
     * @throws ValidationException
     */
    private function logException($index, int $group, Exception $exception): void
    {
        $this->getLogger()->error($exception->getMessage());
        $msg = sprintf(
            'Error indexing core %s on group %s,' . PHP_EOL .
            'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL,
            $index,
            $group
        );
        SolrLogger::logMessage('ERROR', $msg, $index);
    }
}


1		<?php
2
3
4		namespace Firesphere\SolrSearch\Tasks;
5
6		use Exception;
7		use Firesphere\SolrSearch\Factories\DocumentFactory;
8		use Firesphere\SolrSearch\Helpers\SolrLogger;
9		use Firesphere\SolrSearch\Indexes\BaseIndex;
10		use Firesphere\SolrSearch\Services\SolrCoreService;
11		use Firesphere\SolrSearch\States\SiteState;
12		use Firesphere\SolrSearch\Traits\LoggerTrait;
13		use GuzzleHttp\Exception\GuzzleException;
14		use Psr\Log\LoggerInterface;
15		use ReflectionException;
16		use SilverStripe\Control\Controller;
17		use SilverStripe\Control\Director;
18		use SilverStripe\Control\HTTPRequest;
19		use SilverStripe\Core\Injector\Injector;
20		use SilverStripe\Dev\BuildTask;
21		use SilverStripe\ORM\ArrayList;
22		use SilverStripe\ORM\DataList;
23		use SilverStripe\ORM\DataObject;
24		use SilverStripe\ORM\DB;
25		use SilverStripe\ORM\ValidationException;
26		use SilverStripe\Versioned\Versioned;
27
28		/**
29		* Class SolrIndexTask
30		*
31		* @description Index items to Solr through a tasks
32		* @package Firesphere\SolrSearch\Tasks
33		*/
34		class SolrIndexTask extends BuildTask
35		{
36		use LoggerTrait;
37		/**
38		* URLSegment of this task
39		*
40		* @var string
41		*/
42		private static $segment = 'SolrIndexTask';
43		/**
44		* Store the current states for all instances of SiteState
45		*
46		* @var array
47		*/
48		public $currentStates;
49		/**
50		* My name
51		*
52		* @var string
53		*/
54		protected $title = 'Solr Index update';
55		/**
56		* What do I do?
57		*
58		* @var string
59		*/
60		protected $description = 'Add or update documents to an existing Solr core.';
61		/**
62		* Debug mode enabled, default false
63		*
64		* @var bool
65		*/
66		protected $debug = false;
67		/**
68		* Singleton of {@link SolrCoreService}
69		*
70		* @var SolrCoreService
71		*/
72		protected $service;
73
74		/**
75		* @var int
76		*/
77		protected $batchLength = 1;
78
79		/**
80		* SolrIndexTask constructor. Sets up the document factory
81		*
82		* @throws ReflectionException
83		*/
84	14	public function __construct()
85		{
86	14	parent::__construct();
87		// Only index live items.
88		// The old FTS module also indexed Draft items. This is unnecessary
89	14	Versioned::set_reading_mode(Versioned::DEFAULT_MODE);
90		// If versioned is needed, a separate Versioned Search module is required
91	14	$this->setService(Injector::inst()->get(SolrCoreService::class));
92	14	$this->setLogger(Injector::inst()->get(LoggerInterface::class));
93	14	$this->setDebug(Director::isDev() \|\| Director::is_cli());
94	14	$currentStates = SiteState::currentStates();
95	14	SiteState::setDefaultStates($currentStates);
96	14	}
97
98		/**
99		* Set the {@link SolrCoreService}
100		*
101		* @param SolrCoreService $service
102		* @return SolrIndexTask
103		*/
104	14	public function setService(SolrCoreService $service): SolrIndexTask
105		{
106	14	$this->service = $service;
107
108	14	return $this;
109		}
110
111		/**
112		* Set the debug mode
113		*
114		* @param bool $debug
115		* @return SolrIndexTask
116		*/
117	14	public function setDebug(bool $debug): SolrIndexTask
118		{
119	14	$this->debug = $debug;
120
121	14	return $this;
122		}
123
124		/**
125		* Implement this method in the task subclass to
126		* execute via the TaskRunner
127		*
128		* @param HTTPRequest $request
129		* @return int\|bool
130		* @throws Exception
131		* @throws GuzzleException
132		*/
133	13	public function run($request)
134		{
135	13	$start = time();
136	13	$this->getLogger()->info(date('Y-m-d H:i:s'));
137	13	list($vars, $group, $isGroup) = $this->taskSetup($request);
138	13	$groups = 0;
139	13	$indexes = $this->service->getValidIndexes($request->getVar('index'));
140
141	13	foreach ($indexes as $indexName) {
142		/** @var BaseIndex $index */
143	13	$index = Injector::inst()->get($indexName, false);
144
145	13	$indexClasses = $index->getClasses();
146	13	$classes = $this->getClasses($vars, $indexClasses);
147	13	if (!count($classes)) {
148	10	continue;
149		}
150
151	13	$this->clearIndex($vars, $index);
152
153	13	$groups = $this->indexClassForIndex($classes, $isGroup, $index, $group);
154		}
155
156	13	$this->getLogger()->info(date('Y-m-d H:i:s'));
157	13	$this->getLogger()->info(sprintf('Time taken: %s minutes', (time() - $start) / 60));
158
159	13	return $groups;
160		}
161
162		/**
163		* Set up the requirements for this task
164		*
165		* @param HTTPRequest $request
166		* @return array
167		*/
168	13	protected function taskSetup($request): array
169		{
170	13	$vars = $request->getVars();
171	13	$this->debug = $this->debug \|\| isset($vars['debug']);
172	13	$group = $vars['group'] ?? 0;
173	13	$start = $vars['start'] ?? 0;
174	13	$group = ($start > $group) ? $start : $group;
175	13	$isGroup = isset($vars['group']);
176
177	13	return [$vars, $group, $isGroup];
178		}
179
180		/**
181		* get the classes to run for this task execution
182		*
183		* @param $vars
184		* @param array $classes
185		* @return bool\|array
186		*/
187	13	protected function getClasses($vars, array $classes): array
188		{
189	13	if (isset($vars['class'])) {
190	1	return array_intersect($classes, [$vars['class']]);
191		}
192
193	12	return $classes;
194		}
195
196		/**
197		* Clear the given index if a full re-index is needed
198		*
199		* @param $vars
200		* @param BaseIndex $index
201		* @throws Exception
202		*/
203	13	public function clearIndex($vars, BaseIndex $index)
204		{
205	13	if (!empty($vars['clear'])) {
206	1	$this->getLogger()->info(sprintf('Clearing index %s', $index->getIndexName()));
207	1	$this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $index);
208		}
209	13	}
210
211		/**
212		* Index the classes for a specific index
213		*
214		* @param $classes
215		* @param $isGroup
216		* @param BaseIndex $index
217		* @param $group
218		* @return int
219		* @throws Exception
220		* @throws GuzzleException
221		*/
222	13	protected function indexClassForIndex($classes, $isGroup, BaseIndex $index, $group): int
223		{
224	13	$groups = 0;
225	13	foreach ($classes as $class) {
226	13	$groups = $this->indexClass($isGroup, $class, $index, $group);
227		}
228
229	13	return $groups;
230		}
231
232		/**
233		* Index a single class for a given index. {@link static::indexClassForIndex()}
234		*
235		* @param bool $isGroup
236		* @param string $class
237		* @param BaseIndex $index
238		* @param int $group
239		* @return int
240		* @throws GuzzleException
241		* @throws ValidationException
242		*/
243	13	private function indexClass($isGroup, $class, BaseIndex $index, int $group): int
244		{
245	13	$this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName()));
246	13	$this->batchLength = DocumentFactory::config()->get('batchLength');
247	13	$totalGroups = (int)ceil($class::get()->count() / $this->batchLength);
248	13	$cores = SolrCoreService::config()->get('cpucores') ?: 1;
249	13	$groups = $isGroup ? ($group + $cores - 1) : $totalGroups;
250	13	$this->getLogger()->info(sprintf('Total groups %s', $totalGroups));
251		do { // Run from oldest to newest
252		try {
253		// The unittest param is from phpunit.xml.dist, meant to bypass the exit(0) call
254	13	if (function_exists('pcntl_fork') &&
255	13	!Controller::curr()->getRequest()->getVar('unittest')
256		) {
257		$group = $this->spawnChildren($class, $index, $group, $cores, $groups);
258		} else {
259	13	$this->doReindex($group, $class, $index);
260		}
261		} catch (Exception $error) {
262		$this->logException($index->getIndexName(), $group, $error);
263		$group++;
264		continue;
265		}
266	13	$group++;
267	13	} while ($group <= $groups);
268
269	13	return $totalGroups;
270		}
271
272		/**
273		* For each core, spawn a child process that will handle a separate group.
274		* This speeds up indexing through CLI massively.
275		*
276		* @param string $class
277		* @param BaseIndex $index
278		* @param int $group
279		* @param int $cores
280		* @param int $groups
281		* @return int
282		* @throws Exception
283		*/
284		private function spawnChildren($class, BaseIndex $index, int $group, int $cores, int $groups): int
285		{
286		$start = $group;
287		$pids = [];
288		// for each core, start a grouped indexing
289		for ($i = 0; $i < $cores; $i++) {
290		$start = $group + $i;
291		if ($start < $groups) {
292		$pid = pcntl_fork();
293		// PID needs to be pushed before anything else, for some reason
294		$pids[$i] = $pid;
295		$config = DB::getConfig();
296		DB::connect($config);
297		if (!$pid) {
298		$this->doReindex($start, $class, $index, true);
299		}
300		}
301		}
302		// Wait for each child to finish
303		foreach ($pids as $key => $pid) {
304		pcntl_waitpid($pid, $status);
305		if ($status === 0) {
306		unset($pids[$key]);
307		}
308		}
309		$commit = $index->getClient()->createUpdate();
310		$commit->addCommit();
311
312		$index->getClient()->update($commit);
313
314		return $start;
315		}
316
317		/**
318		* Reindex the given group, for each state
319		*
320		* @param int $group
321		* @param string $class
322		* @param BaseIndex $index
323		* @param bool $pcntl
324		* @throws Exception
325		*/
326	13	private function doReindex($group, $class, BaseIndex $index, $pcntl = false)
327		{
328	13	foreach (SiteState::getStates() as $state) {
329	13	if ($state !== 'default' && !empty($state)) {
330		SiteState::withState($state);
331		}
332	13	$this->stateReindex($group, $class, $index);
333		}
334
335	13	SiteState::withState(SiteState::DEFAULT_STATE);
336	13	$this->getLogger()->info(sprintf('Indexed group %s', $group));
337
338	13	if ($pcntl) {
339		exit(0);
		1 ignored issue – show Best Practice introduced 2019-11-24 21:19 UTC by Report Bug Copy Issue Report Using `exit` here is not recommended. In general, usage of exit should be done with care and only when running in a scripting context like a CLI script. Loading history...
340		}
341	13	}
342
343		/**
344		* Index a group of a class for a specific state and index
345		*
346		* @param $group
347		* @param $class
348		* @param BaseIndex $index
349		* @throws Exception
350		*/
351	13	private function stateReindex($group, $class, BaseIndex $index): void
352		{
353		// Generate filtered list of local records
354	13	$baseClass = DataObject::getSchema()->baseDataClass($class);
355		/** @var DataList\|DataObject[] $items */
356	13	$items = DataObject::get($baseClass)
357	13	->sort('ID ASC')
358	13	->limit($this->batchLength, ($group * $this->batchLength));
359	13	if ($items->count()) {
360	1	$this->updateIndex($index, $items);
361		}
362	13	}
363
364		/**
365		* Execute the update on the client
366		*
367		* @param BaseIndex $index
368		* @param $items
369		* @throws Exception
370		*/
371	1	private function updateIndex(BaseIndex $index, $items): void
372		{
373	1	$client = $index->getClient();
374	1	$update = $client->createUpdate();
375	1	$this->service->setInDebugMode($this->debug);
376	1	$this->service->updateIndex($index, $items, $update);
377	1	$client->update($update);
378	1	}
379
380		/**
381		* Log an exception if it happens. Most are catched, these logs are for the developers
382		* to identify problems and fix them.
383		*
384		* @param string $index
385		* @param int $group
386		* @param Exception $exception
387		* @throws GuzzleException
388		* @throws ValidationException
389		*/
390		private function logException($index, int $group, Exception $exception): void
391		{
392		$this->getLogger()->error($exception->getMessage());
393		$msg = sprintf(
394		'Error indexing core %s on group %s,' . PHP_EOL .
395		'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL,
396		$index,
397		$group
398		);
399		SolrLogger::logMessage('ERROR', $msg, $index);
400		}
401		}
402

Firesphere / silverstripe-solr-search

Push — hans/cow-is-animal ( cb2f8e...10edfb )

SolrIndexTask::doReindex() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like